All the same principles apply when using resurrection. It doesn't reduce losses to 0, but only halves them. That is, large losses still remain large.
Essentially, resurrection provides the opportunity to defeat stronger neutral creatures earlier with the same losses. For example, defeating 100 hypothetical Hydras without resurrection may be equivalent in terms of losses to defeating 120 hypothetical Hydras with resurrection.

Plus, even if we accept this point of view, developing in the Castle is already a deliberately chosen strategy, so why should it fit all decisions?

In the first month, the middle, and sometimes even the end of the game is often determined. The obvious endgame is when everything is built, and there is nowhere else to develop, and you simply accumulate troops (the most tedious part).